AGENT SKILL

ros1-operations

Inspect ROS 1 Noetic-or-newer graphs and perform bounded ROSA node, topic, service, parameter, package, log, launch, and node actions with discovery-first safety.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
ros1-operations
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Inspect ROS 1 Noetic-or-newer graphs and perform bounded ROSA node, topic, service, parameter, package, log, launch, and node actions with discovery-first safety.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

ROS 1 Operations

Use this sub-skill when the task concerns ROS 1/Noetic-or-newer, the ROS master, rosnode, rostopic, rosservice, rosparam, ROS package or launch files, ROS logs, or ROS 1 middleware errors. It describes ROSA's package-level tools; it is not a shell-command recipe and must not be treated as a substitute for a sourced ROS installation.

Scope and prerequisites

  • ROSA must be constructed for ROS 1 (for example, the parent agent's ROSA(ros_version=1, ...) path). See agent-core for constructor, provider, executor, and runtime setup.
  • A compatible ROS 1 Python environment must provide the imported ROS modules (rosgraph, rospy, rosnode, rosparam, rospkg, rosservice, rostopic, and rosmsg) and a reachable ROS master for live graph work. Without those prerequisites, report the middleware/import failure; do not infer that the graph is empty.
  • Start with discovery. Never invent a node, topic, service, package, launch file, parameter, or log path. Use the returned names in later calls.
  • Execute exactly one ROSA tool call at a time and wait for its result before selecting the next call. Do not parallelize inspection or actions.
  • Treat rosservice_call, rosparam_set, roslaunch, and rosnode_kill as potentially mutating. Explain the target and expected side effect and obtain confirmation before an action when the user has not already given clear authorization. roslaunch and rosnode_kill are especially high risk.

Operating procedure

  1. Classify the request as graph inspection, bounded observation, read-only metadata lookup, or mutation.
  2. For an action request, first call rosnode_list and rostopic_list with their defaults unless a specific, already-known non-root namespace is the task. For a service action, also call rosservice_list. These calls establish current availability; they do not prove a ROS master exists if they return an error.
  3. Select only names present in the latest result. Before rostopic_echo, rosservice_call, rosparam_set, roslaunch, or rosnode_kill, inspect the selected entity or package with the corresponding info/list tool. For services, inspect service metadata and the service type before constructing arguments. For a launch, enumerate package launch files first.
  4. Apply narrow pattern, namespace, and blacklist filters when an unfiltered list is too large. Use rosgraph_get for connected publisher-topic- subscriber relationships, never as a general node/topic list API.
  5. Bound observations: keep topic echo counts in the documented 1--100 range, use a finite timeout, and avoid returning large message payloads. For logs, discover directories/files before reading a bounded recent slice.
  6. After a mutation, report the exact tool result. Re-list or re-inspect when the result needs confirmation; do not claim a launch, kill, parameter write, or service effect from intent alone.

API and failure details

  • Read api-reference.md for exact tool inputs, defaults, filtering semantics, and response shapes.
  • Read workflows.md for entity discovery, graph, service, parameter, package, launch, log, and sequential-action workflows.
  • Read troubleshooting.md for missing modules or master, empty graphs, namespace/regex behavior, bounded output limits, service arguments, launch risks, blacklist effects, and kill confirmation.

Boundaries and related skills

  • This skill covers ROS 1 APIs in rosa.tools.ros1 plus the shared ROSA read_log behavior needed to consume discovered logs. It does not cover ROS 2 CLI-backed APIs; route ROS 2 requests to ros2-operations.
  • It does not cover generic ROSA construction, model/provider setup, executor lifecycle, or package installation; route those to agent-core.
  • It does not teach custom tool registration, prompt customization, or general blacklist injection; route those to tool-customization.
  • The parent router and shared prerequisites are in rosa. No source checkout, demo script, Docker/X11 setup, or live ROS script is a runtime dependency of this sub-skill.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351